Researchers used a deep-learning pathology model to predict nivolumab outcomes in gastric cancer, building toward pathology-derived risk stratification before immunotherapy begins. The work, published in the British Journal of Cancer, describes a risk score designed to forecast patient response to nivolumab based on tumor microscopic features. If validated prospectively, similar models could change how clinicians select immunotherapy candidates and how trials enrich for responders—particularly in biomarker-underserved settings.